Multiply Enhanced Odd-Order Wave-Mixing ... - ACS Publications

Oct 27, 2009 - Nathan A. Mathew, Stephen B. Block, Lena A. Yurs, Kathryn M. Kornau, Andrei V. ... Department of Chemistry, UniVersity of Wisconsin Mad...
1 downloads 0 Views 2MB Size
13562

J. Phys. Chem. A 2009, 113, 13562–13569

Multiply Enhanced Odd-Order Wave-Mixing Spectroscopy Nathan A. Mathew, Stephen B. Block, Lena A. Yurs, Kathryn M. Kornau, Andrei V. Pakoulev, and John C. Wright* Department of Chemistry, UniVersity of Wisconsin Madison, Madison, Wisconsin 53706 ReceiVed: June 19, 2009; ReVised Manuscript ReceiVed: September 11, 2009

Extending current coherent multidimensional spectroscopy (CMDS) methods to higher order multiwave mixing requires excitation intensities where dynamic Stark effects become important. This paper examines the dynamic Stark effects that occur in mixed frequency/time domain CMDS methods at high excitation intensities in a model system with an isolated vibrational state. The phase-matching restrictions in CMDS define the excitation beams that interact by nonlinear mixing while the dynamic Stark effects create vibrational ladders of increasingly more energetic overtone and combination band states. The excited quantum states form coherences that reemit k1 + b k2 + k2 + b k2′ and b kout )- b the output beams. This paper uses the phase-matching conditions b kout ) b k1 - b b k2′, where the subscripts denote the excitation frequencies of each excitation pulse and the output pulse. The phase-matching condition constrains each pulse to have an odd number of interactions so the overall mixing process that creates the output coherence must also involve an odd number of interactions. Tuning the excitation frequencies and spectrally resolving the output intensity creates three-dimensional spectra that resolve the individual overtone states. Changing the excitation pulse time delays measures the dynamics of the coherences and populations created by the multiple excitations. The multidimensional spectra probe the highly excited states of a molecular potential energy surface. This paper uses tungsten hexacarbonyl (W(CO)6) as a model for observing how dynamic Stark effects change the multidimensional spectra of a simple system. The simplicity of the W(CO)6 system provides the experimental data required to develop the nonperturbative theoretical methods that will be necessary to model this new approach to CMDS. Introduction The selectivity of multiple pulse nuclear magnetic resonance (NMR) rests on its ability to excite multiple spin transitions to create coherent superposition states that are selective for specific molecules.1-3 The multiple transitions must occur within the superposition state’s dephasing time. Similarly, the development of the coherent optical analogues to multiple pulse NMR requires the ability to excite multiple optical transitions within the dephasing time of the coherent superposition states. Since dephasing rates for optical transitions are orders of magnitude faster than spin transitions, high excitation intensities are required to create Rabi frequencies that are comparable to the coherence dephasing rates. High intensities, however, create dynamic Stark effects, so it becomes important to understand how dynamic Stark effects change CMDS spectra. Dynamic Stark effects are commonly used to create the multiple interactions needed to generate highly excited state populations in pump-probe experiments and control reaction dynamics in coherent control experiments.4-7 Chirped pulses5 and temporally shaped pulses7 are particularly effective in driving multiple interactions because the changing frequencies can match the changes in the anharmonicity of increasingly energetic states. In this paper, we use W(CO)6 as a model compound for exploring the effects of dynamic Stark effects on coherent multidimensional spectra. W(CO)6 is a model compound because it has an isolated vibrational transition and a very large transition moment.8-22 Thus, dynamic Stark effects can be obtained with modest excitation intensities and observed without competing effects from overlapping transitions. This paper * To whom correspondence should be addressed. E-mail: wright@ chem.wisc.edu.

describes a mixed frequency/time domain multiply enhanced odd-order wave-mixing (MEOW) approach where multiple beams and dynamic Stark effects coherently create a series of ladder-climbing excitations that populate highly excited states and probe their dynamics. The experiments demonstrate that this method can excite the V ) 6 overtone of the symmetric stretch mode using 12 wave mixing. Three excitation pulses create coherent output beams by MEOW, and a monochromator spectrally resolves the output frequencies. The intensity of the first two excitation pulses is high enough to cause multiple interactions and climb the potential well using the dynamic Stark effects while the last pulse typically has a lower intensity and interacts a single time. Changing the excitation and monochromator frequencies allows measurement of the multidimensional spectra. Changing the time delays between the three excitation pulses allows measurement of the dynamics of the coherences and populations created by MEOW. The dynamic Stark effect broadens the transitions so even overtone states with large anharmonicities can be efficiently excited using excitation sources whose bandwidth would not normally cause a transition. Theory In a phenomenological description of multiwave mixing, the polarization of an isotropic system is written as a Taylor series expansion in the field

P f χ(1)E + χ(3)E3 + χ(5)E5 + · · ·

(1)

since the even terms vanish by symmetry.23 Each χ(n) represents n electric field/matter interactions. Our experiment uses three discrete excitation pulses with two independently tunable

10.1021/jp905796y CCC: $40.75  2009 American Chemical Society Published on Web 10/27/2009

Odd-Order Wave-Mixing Spectroscopy

J. Phys. Chem. A, Vol. 113, No. 48, 2009 13563 excited state populations. For example, one lowest order Liouville pathway that creates the 3a,3a population is 2′

-2′

-2′

2′

-2

2′

ggfagf a,af2a,af 2a,2af3a,2af3a,3a but there are many other equivalent pathways that result in the same population and all must be considered. The last pulse (labeled 1) is weak and interacts only once. It interacts with the 3a,3a population and can create a 4a,3a or 3a,2a output coherence by 1

1

3a,3af4a,3a Figure 1. Liouville pathways for the multiply enhanced odd-wave mixing. The letters designate the density matrices describing the coherences and populations resulting from interaction with an electromagnetic field. The arrows are double headed because in the strong field limit, transitions can occur in either direction. Horizontal arrows correspond to ket-side transitions and vertical arrows correspond to bra-side arrows. Boxes denote the initial ground state population, gg, and the output coherences created with the b k4 ) b k1 - b k2 + b k2′ phasematching condition.

frequencies, ω1 and ω2. The ω2 beam is split to create the third excitation pulse, ω2′. Under these conditions, E ) E1 + E2 + E2′. In coherent experiments, the nonlinear polarization represented by eq 1 is spatially coherent and emits beams in directions determined by the phase-matching condition. This technique allows weak nonlinear signals to be observed against stronger processes which emit in 4π steradians. The direction is determined by momentum conservation as defined by the k-vector addition of the interacting laser fields. The two phase matching conditions for the b kout output beam in our experiments b1 + b are b kout ) b k1 - b k2 + b k2′ and b kout ) -k k2 + b k2′. In a quantum mechanical description of multiwave mixing, Liouville diagrams describe the sequence of coherences and populations created by successive excitation pulses.24 Figure 1 shows all the possible pathways for our MEOW experiment. Here, the arrows designate interactions with the excitation pulses and the letters designate the density matrix elements describing the coherences and populations resulting after each interaction. The first and second letters describe the ket and bra state, respectively, of a |m〉〈n| coherence or population. The boxes indicate either the ground state population (gg) or an emitting output coherence (e.g., 3a,2a). The arrows are not labeled with the fields creating the interactions because the interactions can be caused by different excitation fields. The experiments in this paper use three separate fields, but each field can create multiple interactions. After multiple interactions, fields with positive k vectors cause net ket state changes to move right in Figure 1 or bra state changes to move up. Fields with negative k vectors cause the opposite changes. In the strong field limit, the arrows are double-headed since forward and reverse transitions occur. The transitions created by any one field, however, are constrained by the phase-matching condition to an odd number so the net momentum matches the phase-matching condition. For b2 vector for excitation example, the phase-matching has a -k pulse 2. If it interacts N times, the phase-matching requires b2 ) -k b2 and that N must be b2 - [(N + 1)/(2)]k [(N - 1)/(2)]k odd. For the conditions used in this paper, excitation pulses 2 and 2′ are temporally overlapped and pulse 1 is much weaker and delayed by 6 ps. Since pulses 2 and 2′ have opposite signs in the phase-matching, their net effect is to create a series of transitions from the ground state population to the different

or

3a,3af3a,2a

respectively. Both output coherences result from these seventh order processes and are eight-wave mixing pathways. The 3a,2a output coherence can also be created by a fifth order 6-wave mixing process such as 2′

-2′

2′

-2

1

ggfagf2a,gf 2a,af2a,2af3a,2a Since both processes are quantum mechanically equivalent, the contributions from both must be summed and will interfere. If the excitation frequencies are fully resonant, the successive odd orders have opposite signs, so the interference between successive odd orders is destructive. As a specific example, we write the density matrix elements of the output coherences for the two example Liouville pathways in the steady state limit where the excitation pulses are long compared to the dephasing time. This limit is appropriate for the experiments in this paper. The expressions are

F3a,2a )

2 Ω2agΩ2a,a Ω3a,2a × 32∆ag∆2a,g∆2a,a∆2a,2a∆3a,2a b

b

b

b

Fggei[2k 2-k 2-k 2+k 1)z-(2ω2-ω2-ω2+ω1)t]

F3a,2a

(2)

2 3 Ω2agΩ2a,a Ω3a,2a ) × 64∆ag∆aa∆2a,a∆2a,2a∆3a,2a∆3a,3a∆3a,2a b

b

b

b

Fggei[3k 2-2k 2-k 2+k 1)z-(3ω2-2ω2-ω2+ω1)t]

(3)

for the fifth and seventh order processes, respectively. Here, Ωij ≡ (µijE°)/(p) is the Rabi frequency where E° is the electric field associated with the iTj transition; ∆ij ≡ ωij - ωlasers iΓij is the resonance detuning factor for the ij coherence; ωlasers is the combination of excitation frequencies creating the ij coherence; and µij, ωij, and Γij are the transition moment, frequency, and dephasing rate of the ij coherence, respectively. The k vectors and excitation frequencies from each interaction are shown explicitly to reflect the order of the pathway. On resonance, eqs 2 and 3 become

F3a,2a )

2 iΩ2agΩ2a,a Ω3a,2a b b b F ei[k 2-k 2+k 1)z-ω1t] 32ΓagΓ2a,gΓ2a,aΓ2a,2aΓ3a,2a gg

(4) and

13564

J. Phys. Chem. A, Vol. 113, No. 48, 2009

F3a,2a ) -

2 3 iΩ2agΩ2a,a Ω3a,2a 2 64ΓagΓaaΓ2a,aΓ2a,2aΓ3a,3aΓ3a,2a

b

Mathew et al. b

b

Fggei[k 2-k 2+k 1)z-ω1t]

(5) so the different orders of interactions result in opposite signs for the pathways. In addition to interference between pathways with different orders of interactions that create the same coherence, one must also consider interference between different output coherences arising from the same N-wave mixing process. Successively higher output coherences differ in sign because they involve different numbers of bra-side interactions. Each bra-side interaction changes the sign of the nonlinear polarization. For example, the ag and 3a,2a coherences involve an even number of braside transitions and constructively interfere while they destructively interfere with the 2a,a and 4a,3a coherences, which have an odd number of bra-side interactions. The experiments reported in this paper involve electric fields that are intense enough that dynamic Stark effects become important. The dynamic Stark effect can be conceptualized as a breakdown of the Taylor series perturbative expansion.25,26 As the excitation intensity increases, the higher order coherences and populations become important and the complex interference effects mentioned above come fully into play. Ultimately, the vibrational transitions split and widen. In the steady state limit of a closed two-state system, the dynamic Stark effect causes the following well-known relationship:25-27

train is divided and excites two independently tunable optical parametric amplifiers (OPAs). Difference frequency generation from the OPA signal and idler create tunable mid-IR light. One OPA creates the ω1 beam, and the second OPA output is divided to create the ω2 and ω2′ beams. Typical energies in our experiments range from 0.2 to 1.5 µJ for ω1 and 2.3 and 1.2 µJ for ω2 and ω2′, respectively. The spectral width (fwhm) is 20 cm-1. The spectral output also has two satellite features that are